snisarenko

Not necessarily. I am not an expert on this topic, but I believe physical device failure is one of the least common reasons a packet is dropped (at least on physical cable transmission). I think link saturation, priority routing (for certain types of packets, or contractual reasons) might be more common. There are thousands of routers out there with a variety of different configurations that decide which packets go through under which conditions.

I guess I meant it was an interesting topic to me when I read the paper :) It might not be that exciting overall.

If you are streaming a live sports event you don't want the video frames to fall behind from the real clock too far, but you also don't want the video to stutter when there is packet loss. I believe you can use the UDP trick I mentioned above to amortize loss of high quality frame data, with "double sent" low quality frame data.

I remember reading a paper a while ago that showed that if you send two consecutive UDP packets with exact same data over the internet, at least 1 of them will arrive to the destination at pretty high success rate (something like 99.99%)

I wonder if this still works with current internet infrastructure, and if this trick is still used in real-time streaming protocols.

But if we go back to thinking about this problem from first principles [1], I think certain things have changed.

1. Hosting a complex app in the "cloud" has become significantly cheaper, and easier

2. Centralized services have network effects because you can't find your "connections" on other services. That's not true with federation. Even though gmail is the most popular provider of email, there are still other providers that are alive and well today. In fact there has been some cultural movement of moving from gmail to paid & private email providers, because they are so cheap nowadays.

3. In federated systems, its relatively "easy" to move to another provider. Although much more work needs to be done to improve migration in fediverse.

4. De-centralization doesn't have to be an all or nothing proposition. What is the minimal "centralized", and cheap bootstrapping service for a federated system ? Answer: A domain name + static site with some links to "trusted" providers. Cost: $10 a year for domain + static hosting (pretty much zero dollars) ~ approximately $1 dollar per month

5. There has been some egregious censorship happening by centralized social providers (a total 180 to how early internet culture was). It's their right to do it as private companies, but that's also what motivated some people to move to the fediverse.

[1] https://fs.blog/2018/04/first-principles/

I'll quote it here

> My long term thoughts on aggregators for fediverses:

I think aggregation/curation sites (like pixelfed.club) can be a decentralized complement to the fediverse. The fediverse generates the content. Aggregators, can curate the content for specific topics and help new users discover content. Anyone can start a fediverse instance, and anyone can start a aggregation/curation instance.

Seems like you guys fit the bill. I'll do the 30 day trial, and see how it goes. Thanks for the response.

I am currently using logdna as my splunk alternative, but it's not very good.

Why should I try scalyr next, over the other alternatives ?

My main constraint is the solution should be at least half as good as splunk, and cost less than $50 a month (I generate less than 1G a day).

Your pricing seems to be reasonable. But how do you compare to splunk on features ?

Here are the other tools I am considering trying

https://www.humio.com/

https://www.sumologic.com/

https://sematext.com/logsene/

https://www.datadoghq.com/log-management/

https://logz.io/pricing/

https://www.chaossearch.io/

https://www.papertrail.com/

https://www.loggly.com/

https://www.graylog.org/products/enterprise

You can see the original quality image by clicking on the post. It will take you to the actual post on the actual pixelfed instance.

Yep, that's a good point. The next feature I am planning for mastodonia is to filter by language

> Are you going to release the source code?

Good question. Not sure yet. I am trying to figure out if commercialization is a good path for fediverse related projects. I think fediverse community is pretty anti VC and profit. On the one hand I kind of agree with them - VC money eventually ruins all community/social related startups (ex: digg, reddit, twitter, facebook, instagram etc). It's a cycle that's been repeating for about 20 years (maybe the market is about to realize it and change direction, hence the fediverse). On the other hand you can't maintain fediverse instances on charity alone. I think a good trade-off are non-profit corp structure.

I think we can learn a lot from high quality communities that have survived and remained high signal in the past 20 years.

Some examples:

Hackernews: Subsidized by YCombinator, to maintain an funnel of companies for their seed fund.

https://lobste.rs - Hosting seems to be based on benevolent charity from prgmr.com

somethingawful.com - a forum thats been around for a while and charges $10 to join.

My long term thoughts on aggregators for fediverses:

I think aggregation/curation sites (like pixelfed.club) can be a decentralized complement to the fediverse. The fediverse generates the content. Aggregators, can curate the content for specific topics and help new users discover content. Anyone can start a fediverse instance, and anyone can start a aggregation/curation instance.

I built pixelfed.club to reduce friction between people finding out about pixelfed (and other fediverses) and being able to see whats happening on those Fediverses.

Any feedback is welcome.

I've also built a similar aggregator for Mastodon, you can check it out here:

https://mastodonia.club

(Warning: Some NSFW posts might show up, I do my best to filter them out)

I am working on some fediverse aggregator projects. Let me know if you want to talk.

https://mastodonia.club

https://pixelfed.club

I've seen comments in several previous HN discussions respectfully address the mods. Based on this I assumed it was fair to address comments to mods if it helps improve quality of HN discussion.

> That’s like, just your opinion, man.

It is my opinion that if any siting president (of any party) can abuse FISA court to spy on an opponents presidential campaign, the US democracy is in trouble

> Hungary established dictatorship > See, this one actually did fundamentally change how government works.

Spying on a presidential campaign is in my opinion a fundamental change in how US government works.

There is no dedicated meta-discussion section on hacker-news. So it has to happen somewhere.

And no I am not trolling. I do value the high signal to noise ratio on HN, which is why I participate. My comment is meant to make not just mods, but users think about what counts as signal, versus noise on this forum.

I don't want leave unrelated comments. But I have genuine question here.

Why is this article allowed to stay up on Hacker News and be open for discussion ? How is this related to hacker news or hacker/tech culture ?

Is it worrying that Hugary's PM is grabbing authoritarian powers? Sure. Does this have wide implications for Europe? Sure.

But if this article is allowed to stay up. Then why was the post about the House Intelligence Committee Report on FISA Abuses nuked from the front-page last February ?

https://news.ycombinator.com/item?id=16292336

One would think, illegally spying on a US presidential campaign is pretty worrying as well (its a bi-partisan issue, that could affect both parties, so put down your pitchforks). In fact, such event could undo the underpinnings of the US democracy. A pretty important topic to discuss I'd say.

Required reading

http://www.truthcoin.info/blog/pow-cheapest/

HN Discussion

https://news.ycombinator.com/item?id=16451187

Also newer article http://www.truthcoin.info/blog/pos-still-pointless/

What I learned from talking to people, the only people who have trouble finding weekend activities are travelers, or couples. But I am not sure if those markets are big enough considering the complexity of building this. I have some iteration ideas that i might try in the future. I also discovered existing solutions that are "good enough" (sffuncheap etc.).

Also what does https://www.roast.io/ do better than netlify. I think you should highlight it on your homepage. Thats your main competitor as far as I see it.

Found the HN thread: https://news.ycombinator.com/item?id=22013700

And the related article https://baremetrics.com/blog/i-almost-sold-baremetrics-for-5...

I think the flaw in your reasoning is the "genetic" logical fallacy https://yourlogicalfallacyis.com/genetic

Just because someone "bad" doesn't want something, doesn't make that thing automatically "good"

The problem is that they can't compete on price (in the downwards direction), and they can't compete on building new infra with latest technology. The minimum price is established by the price the owner of the pipes charges. Which is usually set by the gov't. So the can only compete on value added services (i.e. hey we are not going to sell your data hooray!)

The core competition we want in local infrastructure is faster internet and cheaper prices. I think the dual bid system can accomplish that (i.e. separate bids on building/upgrading infra, and bids on maintenance/support)

As for gas and electricity, I think a similar bid system would work as well.